See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332429701

Recommendation System for Smart LMS Using Machine Learning: A

Literature Review

Conference Paper · September 2018

DOI: 10.1109/ICCED.2018.00031

CITATIONS READS

26 676

5 authors, including:

Dina Fitria Murad Yaya Heryadi

Binus University Binus University
75 PUBLICATIONS   274 CITATIONS    106 PUBLICATIONS   729 CITATIONS   

SEE PROFILE SEE PROFILE

Bambang Dw
Binus University
21 PUBLICATIONS   88 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

An Analysis the Impact of Culture Component and Its Effect on Implementation of Knowledge Management View project

Bioinformatics and Biological Computation View project

All content following this page was uploaded by Bambang Dw on 30 June 2019.

The user has requested enhancement of the downloaded file.

 Recommendation System for Smart LMS using
Machine Learning: A Systematic Literature
Review
Dina Fitria Murad Yaya Heryadi Bambang Dwi Wijanarko
1 1 1
Computer Science Department, BINUS Computer Science Department, BINUS Computer Science Department, BINUS
Graduated Program-Doctor of Computer Graduated Program-Doctor of Computer Graduated Program-Doctor of Computer
Science, Bina Nusantara University, Science, Bina Nusantara University, Science, Bina Nusantara University,
Jakarta, 11480, Indonesia Jakarta, 11480, Indonesia Jakarta 11480, Indonesia
2 2
Information Systems yayaheryadi@binus.edu Computer Science
Binus Online Learning Binus Online Learning
Bina Nusantara University Bina Nusantara University
Jakarta, Indonesia Jakarta, 11480, Indonesia
dmurad@binus.edu bwijanarko@binus.edu

Sani Muhamad Isa Widodo Budiharto

1 1
Computer Science Department, BINUS Graduated Program- Computer Science Department, BINUS Graduated Program-
Doctor of Computer Science, Bina Nusantara University, Doctor of Computer Science, Bina Nusantara University,
Jakarta, 11480, Indonesia Jakarta, 11480, Indonesia
sisa@binus.edu wbudiharto@binus.edu

Abstract - This paper presents the result of Systematic Literature mode from conventional or face-to-face learning. This learning
Review (SLR) on Recommender System (RS) topic as a preliminary mode were expected to reduce geographical barier (e.g. the students
toward a further study on designing a smart Learning Management can learn from anywhere) and time barrier (e.g. students can learn in
System (LMS) for online learning which adopts Natural Language their convenient anytime). In contrast to conventional ways of
Processing techniques. As a foundation to a broader study on learning, communication between students and lectures in e-
smart LMS, this study focused on analyzing prominent study learning is not face-to-face but is facilitated by a software called
reports on recommender systems in general and online learning in Learning Management System (LMS). In general, LMS is a web-
particular. The SLR method analyzed papers published in the based platform used to automate and centralize the administration
range of 2013-2018. Out of the 109 papers this study analyzed in- of online learning activities such as registration, learning material
depth 42 papers. The study findings confirmed that most of RS delivery, tracking and reporting on learning progress. An LMS also
studies still focused on e-commerce, movies, tourists, and more
provides functionalities such as: user management, course
whose most popular RS methods were collaborative filtering and
organization, shipping only, online assessment, question handler,
content base. Some studies in RS for online education were mostly
reporting, communications and collaboration services. Effective
focused on scheduling, recommendations for courses, books,
prospective students and others. The results of this study found that LMS must provide a learning environment to learners, so
there are still much opportunities to develop methods and administrators can use to easily manage users and manage other
approaches for RS in online learning. This study findings gives online learning services [1].
foundation of our future research to develop a model of conscious In Indonesia, since 2014 the Government of Indonesia set out an
contextual recommendation system using Machine Learning based online learning initiative for higher education to increase student
on smart LMS for online learning. access to quality courses typically provided by qualified colleges or
lecturers. The particular objectives of this initiative were to address
Keywords: Recommender System, smart LMS, distance several general issues such as: (i) limited access and capacity of
learning, online learning, machine learning. high quality colleges which are mostly concentrated in Java and (ii)
limited quality education resources. The program which was
I. INTRODUCTION initially called Open and Integrated Indonesian Open Learning
Online learning (also known as distance learning or e-learning) (PDITT) later called the Indonesian Online Learning System
is a term refers to training provided via a computer or other digital (SPADA Indonesia). By using the SPADA Indonesia system,
device. In the last ten years, online learning in higher education has students enrolled in any university in Indonesia can take available
been widely accepted in many countries as an alternative learning
online courses from the chosen universities and the credit obtained based on user preferences, interests, or observed behavior about the
will be accepted by the college where the student enrolled. item.
In order to achieve these objectives, the universities as online The study by Kumar [7] on RS for marketing purpose concluded
learning organizers (providers) are required to foresee all necessary that RS is an effective tool to provide recommendations on products
management and technical aspects of online learning services. / services to certain consumer. Google, Amazon, and Netflix
Although looked easy, online learning service is a challenging experiences show some evidences that well-designed RS can give
service. One of the issue is how to design an effective LMS which good and accurate recommendation system to a large number of
provides convenient functionalities for online learning students but customer. Despite many successful experiences, the study by [7]
a reliable back-end system for online learning management. On the also concluded that the remained problems of RS were: (1) cold
other hand, despite its widely used, to the best of our knowledge, start, (2) Scalability of approach, (3) Accuracy of Suggestions, (4)
most of LMS in many universities are designed and operated as a Changing datasets, (5) Impact of context awareness, (6) Loss of
merely learning repository. The LMS users are required to access neighbor transitivity, (7) Sparsity, (8) Privacy issues, (9)
various functionalities to store or access the required information Recommend items in Long tail.
causes learning monitoring becomes inefficient. This condition has Ricci, Rokach and Shapira [4] added some open issues in the
raised a problem namely: how to design an LMS with automated recommendation system namely: (1) scalability with Big Data and
analytic functionalities using textual/voice documents available in real world data, (2) proactive recommendation system, (3)
LMS repository and students-lectures dialog forum as its input in conservation of privacy, (4) diversity of items recommended for
such away that the LMS is not only able to respond to any inquiry target users Integration of user preference long, (5) short term in the
but some information can be pushed automatically to the recipients. process of building a list of recommendations, (6) generic user
Although much research has been done on the recommendation models and cross-domain recommendation systems, (7) distributed
system; but as far as the author's knowledge, most researchers focus recommendation systems that operate on open networks, (8)
on the accuracy of recommendation systems in predicting recommenders that optimize the order of recommendations, (9)
recommendations rather than knowledge acquired by students. In recommenders designed to operate on mobile devices and (10) the
contrast, there is not much research done on the design and context of use.
measurement of learning system whose design embedded with RS.
As part of addressing this issue, this paper presents a preliminary The recent studies on RS by Onah & Sinclair [8] proposed a
study to on a broader research objective which is designing an collaborative screening framework for the design of a
intelligent LMS which adopts the advent of Natural Language recommendation system for online learning in the form of Massive
Processing (NLP) technologies such as recommender system. As Open Online Courses (MOOC). Adomavicius & Tuzhilin proposed
the first step to address that wide issue, this paper presents the the concept of a conscious contextual recommendation system [5].
results from analyzing previous and related studies on the online Dey, Abowd & Salber [10] proposed Context-aware recommender
learning issue. The remaining of this paper is organized as follows. system is a term for recommendation system designed to provide
Chapter 2 describes some related works. Next, chapter 3 described users with recommendations on some items (eg reading material
Research Method followed by Study Result and Discussion in relevant to the topic under discussion) taking into account the
chapter 4 and Conclusion in Chapter 5. context. In this RS type, the term context is defined as "any
information that can be used to characterize an entity's situation.
II. LITERATURE REVIEW Entities may be places, objects, or persons deemed relevant to the
A recommender system or recommendation system (RS) is a interaction between users and applications, including users and self-
term refers to a computer-based system designed to provide applications." In this RS type, an RS can be represented by a
recommendations to the wearer. According to Zaiane [2], a function F:
F: user × item × context → r (1)
recommender system in education context is a system designed to
Where: user be a learning participant, item be the recommended
provide intelligent recommendations to learners (known as item)
object, and the context be the assessment of the user
namely: the steps / processes to be performed, study program,
condition/situation in the ongoing learning process, and r is the
reading materials (courses, books, articles, website, assessment,
recommendation index of the item.
exercises), or sources of information relating to the topic / subject
Adomavicius & Tuzhilin [6][7] divided further the conscious
based to the choices previously made by participants of learning. In
contextual recommendation system into three types according to its
providing such recommendations, an RS employed a statistical
function: (1) Contextual early processing, (2) Post-contextual
model that is trained in a supervised or unsupervised manner with a
filtering, and (3) Contextual modeling. Based on the ability of the
learning algorithm based on input from the previous learning
recommendation system to accommodate the profile and learning
participants.
styles of online learning participants, Belghis-Zadeh concluded that
The study by Kong, Boll & Heuten [3] on RS for vocational
the recommendation system could support the personalization of
education suggested that an RS should emphasize on supporting the
online learning [8]. The results were validated by Hassan &
transfer and knowledge sharing process that facilitates the diversity
Hamada who also concluded that, in order to be more effective in
of objectives of the learning activities. This study followed by [4]
supporting the online learning process, the recommendation system
[5] [6] suggested that RS should not only provide suggestions for
usefully items to users but also serves to filter information filtering
should be designed to be able to provide recommendations to the 2. Direct Science (www.sciencedirect.com)
learning participants according to their own learning style [9]. 3. Springer Link (link.springer.com)
4. Emerald Insight (www.emeraldinsight.com)
III. METHODOLOGY 5. Google Scholars (https://scholar.google.co.id)
In order to conducts a comprehensive literature review of studies 6. Wiley Online Library (onlinelibrary.wiley.com)
on RS utilization in online learning, the main method of this study is 7. ACM Digital Library (dl.acm.org)
8. Elsevier (https://www.elsevier.com/)
systematic literature review (SLR) methodology [10]. The main
The search strings used for all sources comprises of some strings as
focuses of this review were: keyword patterns for paper search
follows:
process, initiate inclusion and exclusion criteria, data mining, and
1. (recommender system OR RS) AND (context-aware OR
analysis of findings to answer research questions. machine learning) AND (recommendation system OR e-
Systematic Literature Review (SLR) is a prominent method learning)
which can be defined loosely as a systematic method for literature 2. (context-awareness recommendation OR personalized
appraisal. The study by Mulrow published in 1994 [11] and Davis learning) AND (issue OR framework)
et al. in 1995 [12][13] were among evidences that SLR has become 3. (content-based AND collaborative filtering) OR
a common method on health care research domain. Kitchenham et (recommender system AND collaborative)) AND
al. [14] defined SLR as a secondary study method to analyze (recommender system OR learning analytic)
previous research which is focusing on finding, critically evaluating 4. (e-learning personalized OR sentiment analysis) AND
and aggregating numerous important of previous research papers (context-awareness OR learning style)
and relevant to the research topic or questions of interest. 5. (recommender system OR machine learning) AND
Despite different definition, the common objective of this (context-aware OR e-learning)
methodology is to ensure that the review is unbiased, rigorous and The search exclusion criteria in this study were namely:
auditable. Although previously practiced in evidence-based 1. This paper is based on the date of their publication
medicine area, in the past several years, this method has widely between 2013-2018
adopted in various research domains including computer science, 2. The complete paper structure, which means all identities
economics, psychology, social science and health care. The (journal / conference, author's identity, etc.) are
advantage of this method, among others, are to facilitate rational mentioned on paper.
decision making, selecting a large number of information, and keep Duplicate paper from the same study will not be included and
removed from the SLR Data Extractions.
up to date with evidence accumulating in field and to practice
evidence-based medicine.
IV. RESULT
A. Publishing outlets
The source of publication for this study can be summarized in
the following table categorized by its title, paper type and year of
publication.
Table. 1. Source of Publication
Title Year Type Title Year Type
Personalized E- A General Framew
2017 P 2017 J
library: (16)
An Analysis of A Survey on
2016 J 2014 J
Social (17) Method(18)
Toward a New A tag based learning
2015 J 2013 J
(17) (19)
Recommender Exploratory Exam
2015 J 2013 J
sys(6) (20)
Fig.1. Process flow of systematic literature review (source:[15]) A Survey on Reliability Quality
2014 J 2018 J
In this study, SLR was specifically aimed to investigate the Linked(21) (22)
TF -ID uF: A
extent to which the use of RS on online learning education and Novel(23)
2017 J Improvisation of (1) 2017 J
what methods are most frequent and appropriate to use. The current Designing and Exp
2017 J A recommender (25) 2017 J
trend, the high interest of writers to continue to explore the RS to (24)
online learning can identify the different models, methods and A Multi-Stage Recommender Sy
2017 J 2018 J
Appr (26) (27)
approaches used. Based on this, this research will be identified as a
Recommender Sys Designing and Exp
new model in the utilization of RS for online learning. 2017 J 2017 J
(28) (24)
Following et al., [10], the SLR method used in this study PLORS: a
2016 J PCRS: Personalized 2018 J
undertook an automated search in internet-based digital sources and personalized (8)
indexing services including: Collaborative A Sentiment
2015 J 2017 J
Filtering (29) Analysis (30)
1. IEEE Xplore Digital Library (http: /ieeexplore.ieee.org)
 Improvisation of Information Science, Ekiti State University, Federal University of
2017 J Recommender (6) 2014 J
lear (1) Technology, University of California, and University of Ibadan by
Incorporating Proa Recommender Sys
(31)
2013 J
(6)
2015 J publishing 2 papers. Other institutions were only produced one
Implementation of paper each.
2016 J Recommendation(33) 2015 J
(32) In terms of authors' college affiliation, Delhi Technological
Performance and Towards an Intell University published more than 1 paper followed by Department of
2015 J 2017 J
Qu (34) (35)
Computer and Information Science, Ekiti State University, Federal
A Framework for A Personalized E
2017 J 2016 J University of Technology, University of California and University
(36) (18)
Towards Guided of Ibadan. Likewise, this university sequence also related to the
2014 J Recommender (38) 2016 J
(37) number of authors who published more than 1 paper.
Recommendation
2016 J A Multi-Stage (26) 2017 J
(39) D. Publication trends
Academic Decision
(40)
2016 J Open Problems in(4) 2017 J Frequency of papers published between 2013-2018 can be
A privacy- A Framework for seen in the figure. 1. There was a significant increase in RS
2018 J 2017 J
preserving (41) (36) publication. Most publications in 2017 (17 papers) and 2016 (8
Stakeholders
2018 J Learning Object (43) 2016 J papers). And by 2018, as of April 2018 we have found 6 papers on
Analysis(42) RS for this education.
Total 42
E. Authors Country
Note: J be journal and P be proceeding.
As can be seen from Table. 1, it can be concluded that there is Figure. 2 informs the author's home country. This table
a significant increase in the interest of researchers in RS especially provides information on the origin of the colleges of the authors. It
in the context of education, shifting from ecommerce to education. appears that the average author who published more than 1 paper
With the publication of their papers in reputable journals it is came from India RS [34][18][1][44][33][38][28],
known that there is still considerable interest for researchers to USA[26][20][26], Spain[31][40][22][45], Palestine[46],
continue to explore their research in this field. Of course, this Canada[8][36] and Indonesia [47].
becomes an opportunity for the development of appropriate models
and references to hospitals with e-learning especially distance
learning systems.
In this study, SLR was used to analyze 109 papers from all
sources and criteria. Out of those papers, based on titles and
abstracts related to research questions, 55 papers were selected as
study candidates. After further study, there are only 42 papers that
can be used for further analysis due to its strong correlation with
this research as seen in Table 2.
Table 2. Number of papers from selected sources
Source Founded Candidates Selected
Studies Studies Studies
Fig. 2. Distribution of publication year
IEEE 20 2 15 Based on Fig. 3, it can be concluded that India is the country
Science Direct 16 5 1 with the highest level of research in RS research. Particularly in
Springer 12 10 7 some papers, the focus of their research is on e-learning RS
Emerald 8 2 0
[34][18][1][44][33][38] , library and utilization of RS on big data
Google 22 14 10
Schoolar 8 6 0 [28]. How about Indonesia? The government of Indonesia through
ACM 10 6 0 Kemenristek DIKTI has opened up opportunities for the
Wiley 13 10 9 implementation of distance education that will be controlled
Elsevier 109 55 42
directly by the government. This is certainly an opportunity as well
Total
as a challenge to organize a good distance education and quality.
B. Most prolific authors Each university in Indonesia will standardize each other while
As can be seen from the perspective of the author's analysis, improving their quality and quality in various sectors of education.
there are 163 authors who have written 42 papers. There are 29 LMS as a standard media in the implementation of this remote
authors (18%) who are consistent in writing about RS on education. learning system should be maximized with the use of technology,
Other authors on average write only 1 paper in the 2013-2018 such as a recommendation system that will help the education
timeframe. Interestingly, publication numbers in total increased provider in managing its learning and we call it smart LMS using
significantly in the past 5 years (2013-2018) RS with a context-aware approach.

C. Most productive institution F. Authors academic background

Most sources of papers that we used were published in Most RS papers was written by researchers from computer
indexed journals. The most productive institutions are Delhi Science, information technology and information systems fields.
Technological University, Department of Computer and That is, the researchers have the ability in the field of information
technology and this of course helps researchers to further explore Table. 4. Issues in the RS domain research
the utilization of technology especially recommendation system Collabora Content- Demogra Hybrid Domain Knowled
Challenges
tive RS Base RS phic RS RS Base ge Base
that can then be developed with other technologies such as machine Cold Start Problem ? √ √ √ ? √
learning. Interestingly year 2017 is the peak year for RS research in Scalability of √ √ √ ? X X
the education domain although RS research in other areas such as Big Data √ √ √ ? X X
Privacy Concern √ X √ √ √ √
beauty, tourists, health, transportation and others are still continue. Sparsity X √ X √ ? ?
Recommending The
? √ √ √ ? √
Item in The Long Tail
Accuracy of The
? √ X ? X X
Suggestions
Changing Data Set √ √ X √ √ ?
Impact of Context-
? √ ? √ √ ?
Awareness

H. Methods already in use

Of the 42 papers that we have studied, there are several
methods used by previous researchers on the recommendation
system on online learning, ie Collaborative-filtering, Content-
based, Demographic, Utility-based, Knowledge-based,
Figure. 3. Country of author origin
Community-based, Hybrid-approach (mix method). from several
G. Keywords analysis methods and approaches it is known that the most frequently used
Table. 3 informs the keywords used by the authors in their is the method of content-based and filtering collaborative. Based on
papers. This keyword was found on 42 papers in eight sources. that, in our next study we will conduct an experimental test from a
This data shows that the most commonly used keywords are data set that will test the same methods and algorithms and then
"recommender system", "collaborative filtering" and “online compare the results. Previous research by these researchers will be
learning/e-learning". We also found other keywords among the a reference for our next research. From several methods and
papers namely, "Learning Analytic", "data mining", "hybrid", approaches it is known that the most frequently used is the method
"sentiment analysis", "MOOC", but we also find interesting of content-based and filtering collaborative. Based on that, in our
keywords to learn more is the word "internet of thing". next study we will conduct an experimental test from a data set that
This explains that there is still a great opportunity to develop a will test the same methods and algorithms and then compare the
recommendation system suitable for education especially distance results. Previous research by these researchers will be a reference
learning. for our next research.
Table. 3. Most frequently used keywords
#papers V. CONCLUSION
Scienc Recommender System (RS) has become a revolutionary
Keywords Sprin Google Elsevie
IEEE e
ger Scholar r concept facing the big data era. Recommendations in the form of
Direct
E-Learning 2 0 5 3 0 advice for users, is very useful to support achieving student using
recommender systems 15 1 6 5 9 online learning environment. Many of the techniques provided by
Machine Learning 1 0 0 0 0 the recommendation system used of collaborative filtering and
Collaborative filtering 2 1 0 6 6
Content based 1 1 2 1 5 content-based filtering. For online learning it is possible to use both
Recommendation techniques and also to use other techniques such as hybrid,
1 0 0 4 0
system knowledge-based and so on. This study results showed that
Context-aware 1 0 0 0 0 research RS still gained research interest for various research
Learning style 0 2 0 0 0
communities.
SLR results, the three most commonly used RS methods are Interestingly, the SLR results pointed some research
collaborative filtering, content-based and hybrid. Whilst, the most opportunities in RS for online learning domain including:
frequently used approaches in RS are: learning analytic, sentiment 1) Collaborative filtering methods: cold start, recommendation
analysis, data mining and utilization RS on big data. This is in line for item in long tail, and impact of context-awareness.
with the research of Nitin and Kumar [4] who identified Issues in 2) Demographic RS: impact of context-awareness.
RS domain research as show in table. 4. 3) Hybrid RS: Scalability and accuracy of recommendations.
Based on the table 4 it is known that there are still 4) Domain based RS: cold start, recommendation for item in
opportunities to further develop the RS on distance learning using long tail, and sparsity.
machine learning, there are still some methods that can be explored 5) Knowledge based RS: Sparsity, changing dataset, and impact
further utilization. of context awareness.
 [23] J. Beel, S. Langer, and B. Gipp, “TF-ID u F : A Novel Term-Weighting
ACKNOWLEDGEMENTS Scheme for User Modeling based on User s ’ Personal Document
Researchers would like to thank Bina Nusantara University, Collections,” pp. 1–8, 2016.
[24] T. J. Ramabu and H. J. G. Oberholzer, “Designing and Exploring Study
especially BINUS Online Learning, which has supported this Field Recommender System for Prospective Students,” pp. 1–8, 2017.
research, so that research can be done well. Hopefully, this study [25] F. Pesántez-Avilés, D. Calle-López, V. Robles-Bykbaev, M. Rodas-
Tobar, and C. Vásquez-Vásquez, “A Recommender System Based on
can provide more benefits for all. Data Mining Techniques to Support the Automatic Assignment of
Courses to Teachers in Higher Education,” 2017 Int. Conf. Inf. Syst.
REFERENCES Comput. Sci., pp. 231–236, 2017.
[26] T. Morrow, A. R. Hurson, and S. S. Sarvestani, “A Multi-Stage
[1] U. K. Mothukuri et al., “Improvisation of learning experience using Approach to Personalized Course Selection and Scheduling,” 2017.
Learning Analytics in eLearning,” 2017. [27] P. Pilloni, L. Piras, S. Carta, G. Fenu, F. Mulas, and L. Boratto,
[2] O. R. Za, “Building a Recommender Agent for e-Learning Systems,” “Recommender System Lets Coaches Identify and Help Athletes Who
2002. Begin Losing Motivation,” Computer (Long. Beach. Calif)., vol. 51, no.
[3] X. Kong, S. Boll, and W. Heuten, “Towards Recommender Systems 3, pp. 36–42, 2018.
Supporting Knowledge Sharing and Transfer in Vocational Education [28] S. Dwivedi and K. R. Vs, “Recommender System for Big Data in
and Training,” pp. 25–30, 2013. Education,” vol. 2, 2017.
[4] A. Kumar, “Open Problems in Recommender Systems Diversity,” pp. [29] J. E. Sinclair, “COLLABORATIVE FILTERING
82–87, 2017. RECOMMENDATION SYSTEM : A FRAMEWORK IN MASSIVE
[5] Y. Salman, A. Abu-Issa, I. Tumar, and Y. Hassouneh, “A model and OPEN ONLINE COURSES.”
prototype of a proactive multi-type context-aware recommender [30] S. Rani and P. Kumar, “A sentiment Analysis System to Improve
system,” 2017 Intell. Syst. Conf., no. September, pp. 297–303, 2017. Teaching and Learning,” IEEE Comput. Soc., vol. 50, no. 5, pp. 36–43,
[6] A. Klašnja-mili, “Recommender systems in e-learning environments : a 2017.
survey of the state-of-the-art and possible extensions,” 2015. [31] D. Gallego, E. Barra, P. Rodr, and G. Huecas, “Incorporating Proactivity
[7] G. Adomavicius and A. Tuzhilin, “Context-aware recommender to Context-Aware Recommender Systems for E-Learning.”
systems,” Recomm. Syst. Handbook, Second Ed., pp. 191–226, 2015. [32] J. Jooa, S. Bangb, and G. Parka, “Implementation of a Recommendation
[8] H. Imran, S. Graf, and M. B. T. C. Kinshuk, “PLORS : a personalized System using Association Rules and Collaborative Filtering,” Procedia -
learning object recommender system,” Vietnam J. Comput. Sci., vol. 3, Procedia Comput. Sci., vol. 91, no. Itqm 2016, pp. 944–952, 2019.
no. 1, pp. 3–13, 2016. [33] V. Senthil, “Recommendation System for Adaptive E-learning using
[9] M. Hassan, “Smart Media-based Context-aware Recommender Systems Semantic Net,” vol. 63, no. 7, pp. 19–24, 2013.
for Learning : A Conceptual Framework,” 2017. [34] S. Bagchi, “Performance and Quality Assessment of Similarity Measures
[10] B. A. Kitchenham et al., “Refining the systematic literature review in Collaborative Filtering Using Mahout,” Procedia - Procedia Comput.
process-two participant-observer case studies,” Empir. Softw. Eng., vol. Sci., vol. 50, pp. 229–234, 2015.
15, no. 6, pp. 618–653, 2010. [35] M. El Mabrouk, S. Gaou, and M. K. Rtili, “Towards an Intelligent
[11] C. D. Mulrow, “Systematic Reviews: Rationale for systematic reviews,” Hybrid Recommendation System for E-Learning Platforms Using Data
Bmj, vol. 309, no. 6954, p. 597, 1994. Mining,” pp. 52–76.
[12] H. B. Davis D, Andrew M, Oxman A, “Changing Physician [36] M. M. Rahman and N. Aniza, “A Framework for Designing a
Performance : A Systematic Review of the Effect of Continuing Medical Personalised Web-based Search Assistant Tool for eLearning,” vol. 0,
Education Strategies,” Jama, vol. 274, no. 9, pp. 700–705, 1995. no. c, pp. 1–6, 2017.
[13] A. D. Oxman, M. A. Thomson, D. A. Davis, and B. Haynes, “No magic [37] Z. A. Shaikh and S. A. Khoja, “Towards Guided Personal Learning
bullets: a systematic review of 102 trials of interventions to improve Environments,” no. Figure 2, pp. 782–784, 2014.
professional practice,” Cmaj, vol. 153, no. 10, pp. 1423–1431, 1995. [38] M. Dhanda and V. Verma, “Recommender System for Academic
[14] B. Kitchenham, O. Pearl Brereton, D. Budgen, M. Turner, J. Bailey, and Literature with Incremental Dataset,” Procedia Comput. Sci., vol. 89, pp.
S. Linkman, “Systematic literature reviews in software engineering - A 483–491, 2016.
systematic literature review,” Inf. Softw. Technol., vol. 51, no. 1, pp. 7– [39] F. O. Isinkaye, “Recommendation systems : Principles , methods and
15, 2009. evaluation,” Egypt. INFORMATICS J., 2015.
[15] Klassen et al., “Guides for reading and interpreting systematic reviews: [40] Y. Vanessa Niet, V. G. Diaz, and C. E. Montenegro, “Academic decision
I. Getting started.,” Arch. Pediatr. Adolesc. Med., vol. 152, no. 7, pp. making model for higher education institutions using learning analytics,”
700–704, 1998. 2016 4th Int. Symp. Comput. Bus. Intell., pp. 27–32, 2016.
[16] J. Aguilar, P. Valdiviezo-dı, and G. Riofrio, “ORIGINAL ARTICLE A [41] K. Xu, W. Zhang, and Z. Yan, “A privacy-preserving mobile application
general framework for intelligent recommender systems,” pp. 147–160, recommender system based on trust evaluation,” J. Comput. Sci., vol. 26,
2017. pp. 87–107, 2018.
[17] M. D. Lytras, H. I. Mathkour, H. Abdalla, W. Al-Halabi, C. Yanez- [42] A. H. Afridi, “Stakeholders Analysis for Serendipitous Recommenders
Marquez, and S. W. M. Siqueira, “An emerging - Social and emerging system in Learning Environments,” 9th Int. Conf. Ambient Syst.
computing enabled philosophical paradigm for collaborative learning Networks Technol. (ANT 2018), vol. 130, pp. 222–230, 2018.
systems: Toward high effective next generation learning systems for the [43] S. Sergis and D. G. Sampson, “Learning Object Recommendations for
knowledge society,” Comput. Human Behav., vol. 51, pp. 557–561, Teachers Based on Elicited ICT Competence Profiles,” IEEE Trans.
2015. Learn. Technol., vol. 9, no. 1, pp. 67–80, 2016.
[18] K. Anandakumar, K. Rathipriya, and A. Bharathi, “A Survey on [44] G. Article, “Recommender Systems - Comparison of Content-based
Methodologies for Personalized E-learning Recommender Systems,” pp. Filtering and Collaborative Filtering,” vol. 4, no. 5, pp. 3131–3133,
4738–4743, 2014. 2014.
[19] J. Chen, M. Chen, and Y. S. Sun, “Computers & Education A tag based [45] D. Horowitz, D. Contreras, and M. Salamó, “EventAware: A mobile
learning approach to knowledge acquisition for constructing prior recommender system for events,” Pattern Recognit. Lett., vol. 105, pp.
knowledge and enhancing student reading comprehension,” Comput. 121–134, 2018.
Educ., vol. 70, pp. 256–268, 2014. [46] A. Abu-issa et al., “A Smart City Mobile Application for Multitype ,
[20] S. M. Cox, “Exploratory Examination of Relationships between Proactive , and Context-Aware Recommender System,” Int. Conf. Eng.
Learning Styles and Learner Satisfaction in Different Course Delivery Technol., 2017.
Types,” vol. 1, no. 1, pp. 64–76, 2013. [47] ’. I Hariadi and D. Nurjanah, “Hybrid attribute and personality based
[21] S. Dietze, H. Drachsler, and D. Giordano, “A survey on linked data and recommender system for book recommendation,” 2017 Int. Conf. Data
the social web as facilitators for TEL recommender systems,” Recomm. Softw. Eng., no. 2, pp. 1–5, 2017.
Syst. Technol. Enhanc. Learn. Res. Trends Appl., pp. 47–75, 2014.
[22] J. Bobadilla, A. Gutiérreza, F. Ortegab, and B. Zhuc, “Reliability quality
measures for recommender systems,” Inf. Sci. (Ny)., vol. 442–443, pp.
145–157, 2018.

View publication stats